Does macroeconomics help in predicting stock markets volatility comovements? A nonlinear approach

This paper addresses the question of the relevance of macroeconomic determinants in forecasting the evolution of stock markets volatilities and co-volatilities. Our approach combines the Cholesky decomposition of the covariance matrix with the use of the Vector Logistic Smooth Transition Autoregressive Model. The model includes predetermined variables and takes into account the asymmetries in volatility process. Structural breaks and nonlinearity tests are also implemented to determine the number of regimes and to identify the transition variables. The model is applied to realized volatility of stock indices of several countries in order to evaluate the role of economic variables in predicting the future evolution of conditional covariances. Our results show that the forecast accuracy of our model is significantly de m the accuracy of the forecasts obtained via other standard approaches.